Falling and rotating object.

apaojo-5 · February 11, 2010, 07:17

Hi guys!
We have tried to go thru all relevant threads prior to posting this one. Our problem consists of simulating a falling and rotating object. We would like to start with a steady-state case since we aren’t familiar with transient and/or deforming mesh simulations. Our geometry consists of the object being immersed in a cylindrical surrounding, see figure.

http://www.student.ltu.se/~apaojo-5/seed.JPG

So far we have tried setting boundary conditions on both the inlet and the cylindrical wall to axial and theta velocity components, we have also tired to set the cylinder wall to free slip conditions, but this yielded approximately the same results. The problems encountered were that the theta velocity component diminished close to the outlet.

We have also tried the rotating frame approach with an axial component at the inlet, but this resulted in backflow at the outlet and the air being pushed outwards from the center of rotation.

Does anyone have any suggestions on how to solve these problems or how to set up a transient/deforming mesh simulation?

ghorrocks · February 11, 2010, 19:41

You have not explained what you are trying to do. Is your aim to predict the path of a falling object of some shape? Do you expect it to rotate and translate during falling? Does the object deform or is it rigid?

What is the Re number of the object as it falls? Is it falling in air? Do you want the initial transient of just the terminal velocity part?

apaojo-5 · February 12, 2010, 09:35

Hi Glenn,
thanks for shown interest. We are studying the flow around a falling autorotating maple seed in air and the aim is to find the pressure distribution and visualize the assumed leading-edge vortex on the top side of the seed. Thru experiments we have found the terminal descent and rotational velocity, therefore the initial transient part is not of interest at the moment. As a first approximation we assume that the seed is falling straight down with uniform descent and rotational velocity, hence we try to model the seed as stationary and that the surrounding air as moving. The seed is rigid.
Does this further explain our problem?
Thankful for replies!

ghorrocks · February 13, 2010, 06:57

OK, I see now. In that case I would start the analysis as a simple rotating frame of reference simulation, with the rotation rate set to the measured rotation rate. Blow air over it at the known terminal velocity. This should give you a good idea of the flow field at terminal velocity. You can also check it is behaving itself by checking the drag equals the mass of the seed. This analysis is quite straight forward.

The next step would be to link the rotational speed to the torque of the seed so it can adjust the speed of rotation. This is still a RFR model so is quite simple but a bit trickier as the rotation speed now also requires convergence.

The final step, and this is a big step, is to go full 6-DOF solver (or at least several-DOF as you might not need all 6 of the DOF's). This will allow the seed to rock and roll any way it likes but the simulation is significantly more difficult. The simplest approach here is with the immersed grid method, and the moving mesh approach will capture the seed boundary layers better but will be fiendishly difficult to get working. All of these approaches will require the 6-DOF solver which is a beta feature in V12. Contact CFX support to get access to this and some examples.

apaojo-5 · February 25, 2010, 03:39

Thanks Glenn! We have tried the first part and its working! Thanks!

February 11, 2010, 07:17	Falling and rotating object.	#1
apaojo-5 New Member Patrick Jonsson Join Date: Jan 2010 Posts: 5 Rep Power: 16	Hi guys! We have tried to go thru all relevant threads prior to posting this one. Our problem consists of simulating a falling and rotating object. We would like to start with a steady-state case since we aren’t familiar with transient and/or deforming mesh simulations. Our geometry consists of the object being immersed in a cylindrical surrounding, see figure. http://www.student.ltu.se/~apaojo-5/seed.JPG So far we have tried setting boundary conditions on both the inlet and the cylindrical wall to axial and theta velocity components, we have also tired to set the cylinder wall to free slip conditions, but this yielded approximately the same results. The problems encountered were that the theta velocity component diminished close to the outlet. We have also tried the rotating frame approach with an axial component at the inlet, but this resulted in backflow at the outlet and the air being pushed outwards from the center of rotation. Does anyone have any suggestions on how to solve these problems or how to set up a transient/deforming mesh simulation?

February 11, 2010, 19:41		#2
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,854 Rep Power: 144	You have not explained what you are trying to do. Is your aim to predict the path of a falling object of some shape? Do you expect it to rotate and translate during falling? Does the object deform or is it rigid? What is the Re number of the object as it falls? Is it falling in air? Do you want the initial transient of just the terminal velocity part?

February 12, 2010, 09:35		#3
apaojo-5 New Member Patrick Jonsson Join Date: Jan 2010 Posts: 5 Rep Power: 16	Hi Glenn, thanks for shown interest. We are studying the flow around a falling autorotating maple seed in air and the aim is to find the pressure distribution and visualize the assumed leading-edge vortex on the top side of the seed. Thru experiments we have found the terminal descent and rotational velocity, therefore the initial transient part is not of interest at the moment. As a first approximation we assume that the seed is falling straight down with uniform descent and rotational velocity, hence we try to model the seed as stationary and that the surrounding air as moving. The seed is rigid. Does this further explain our problem? Thankful for replies!

February 13, 2010, 06:57		#4
ghorrocks Super Moderator Glenn Horrocks Join Date: Mar 2009 Location: Sydney, Australia Posts: 17,854 Rep Power: 144	OK, I see now. In that case I would start the analysis as a simple rotating frame of reference simulation, with the rotation rate set to the measured rotation rate. Blow air over it at the known terminal velocity. This should give you a good idea of the flow field at terminal velocity. You can also check it is behaving itself by checking the drag equals the mass of the seed. This analysis is quite straight forward. The next step would be to link the rotational speed to the torque of the seed so it can adjust the speed of rotation. This is still a RFR model so is quite simple but a bit trickier as the rotation speed now also requires convergence. The final step, and this is a big step, is to go full 6-DOF solver (or at least several-DOF as you might not need all 6 of the DOF's). This will allow the seed to rock and roll any way it likes but the simulation is significantly more difficult. The simplest approach here is with the immersed grid method, and the moving mesh approach will capture the seed boundary layers better but will be fiendishly difficult to get working. All of these approaches will require the 6-DOF solver which is a beta feature in V12. Contact CFX support to get access to this and some examples.

February 25, 2010, 03:39		#5
apaojo-5 New Member Patrick Jonsson Join Date: Jan 2010 Posts: 5 Rep Power: 16	Thanks Glenn! We have tried the first part and its working! Thanks!